Driving Fatigue Detection Based on EEG Signal

Driving fatigue detection is an important approach to ensure the traffic safety. However, the most existing mature analysis methods are based on driving behavior or driver´s body characteristics, which leads to the low accuracy and predictability. The EEG signal analysis is proved to an effective way to reflect the fatigue state in medical science, thus this paper explores the EEG signal to detect the driving fatigue. We design a portable EEG acquisition system, which detects the drivers´ EEG signals and handles the interference by the median filter, band stop filter and Hilbert-Huang transform. The eigen values are extracted by percentage power spectral density. Two methods are proposed to determine the fatigue levels. Experiment results show that the method based on eigenvalue ratio in eyes-open state has 79% accuracy, the method based on BP neural network in fatigue classification has 83% accuracy, and the eyes-close state recognition rate is more than 97%.